SLR parser

The SLR parser (Simple LR parser) is the most basic member of the LR parser family, introduced by Frank DeRemer in 1969 as a simplification of Donald Knuth's original LR(1) construction. An SLR parser uses a single parser table whose actions are determined by the grammar's LR(0) items and the FOLLOW sets of non-terminals. This makes SLR table construction computationally cheap — far cheaper than canonical LR(1) — but it comes at a cost in grammatic coverage. Many practical grammars that are not ambiguous nevertheless produce conflicts in an SLR table because SLR's lookahead mechanism is too coarse: it treats all occurrences of a non-terminal as having the same FOLLOW set, ignoring the specific context in which each occurrence appears. When SLR fails, the compiler writer typically upgrades to LALR(1), which retains most of SLR's efficiency while using a more precise context-dependent lookahead. SLR's historical importance is not that it is used directly in modern compilers — it rarely is — but that it established the template of table-driven bottom-up parsing that LR, LALR, and GLR parsers all extend.